v3.5.9-beta

Critical bugfix release:

• The Dockerfile was not working. It is now functional.

v3.5.8-beta

• The format keyword is no longer mandatory on ontology YML configuration files. When not present, OSP-core will try to guess the format from the file extension. If such guess fails though, it will still have to be provided.
• OSP-core now uses rdflib >= 6.0.0, < 7.0.0 for users who have python >= 3.7 installed. This translates to performance improvements in some cases, see the benchmarks page. For users who have python < 3.7 installed, rdflib==5.0.0 is still used, as the newer version requires python >= 3.7. This implies that at some point in the medium term, OSP-core's support for Python 3.6 will be dropped.
• Other minor changes (see the full changelog).

v3.5.7-beta

Critical bugfix release:

• rdflib-jsonld has been tombstoned. The latest version 0.6.2 removes all functionality. The developers recommend to use 0.5.0, but it cannot be installed with setuptools >= 58. Luckily, version 0.6.1 is still functional and compatible with setuptools >= 58, so this is the version required now by OSP-core.
• websockets 10 has been released, introducing breaking changes and dropping support for Python 3.6. Thus, for now the requirement websockets < 10 is enforced.

v3.5.6-beta

Critical bugfix release:

• Fixed error affecting default, mutable attributes (issue #679).
• Fixed error when installing ontologies on Windows (issue #684).

v3.5.5-beta

Critical bugfix release:

• rdflib==6.0.0 has been released, but osp-core==3.5.4 is incompatible with it. The python package manager pip was installing rdflib==6.0.0 by default, making new installs unusable. This has now been fixed.
• Fixed regression that broke the ontology2dot tool.

v3.5.4-beta

REMARK: This release is incompatible with rdflib==6.0.0, which was released just a few days afterwards (see issue #677). Right now, the simplest solution is pip install rdflib==5.0.0.

• Added benchmarks, integrated with the CI, which publishes the results automatically here.
• Support for OWL hasValue property restrictions. The values are not enforced, but are set as defaults for new CUDS's attributes.
• Fixed bug when removing CUDS objects directly connected to the wrapper object (they were not removed).
• Added a new type conversion feature to the SPARQL queries. By default, SparqlResult (result) objects return SparqlBindingSet (row) objects when iterated, that in turn return rdflib's URIRefs or Literals when queried (e.g. row['queriedvariable']). Now SparqlResult objects are callable, like result(irivariable='cuds', literalvariable=float), where each keyword argument can be either 'cuds' or a callable. This makes the SparqlResult object return SparqlBindingSet objects that return CUDS objects or python objects when queried, making post-query operations simpler, such as getting a dataframe from the query results.
• Files fetched by the transport session are no longer prepended with the CUDS object's uid. The uid will be added within parentheses after the name, only if there is a name conflict.
• Fixed bug that showed wrong values for the path attribute of file CUDS.
• Other changes and bug fixes (see the full changelog).

v3.5.3.1-beta

• Fixed a bug in the pico ontology installation tool that showed the namespaces xml, rdf, rdfs, xsd, owl as installed even though they were not.
• CUDS objects (ontology individuals) can now have a custom IRI, instead of one from the OSP-core namespace. Just use the iri keyword argument when calling the ontology class to create the CUDS object, e.g. city.Citizen(name='Klaus', age=55, iri='http://example.org/London#Klaus'). However, in order to achieve optimal performance it is recommended to continue using the OSP-core namespace, as in such case OSP-core only has to store and handle an uuid, instead of an iri, which is usually far longer.
• The namespace objects now support IPython tab completion (click here to see an example). See the docs for more details.
• Added basic SPARQL support. It works on the core session (although with performance issues specific to the core session) and on all triple store wrapper sessions (it's mandatory to implement the support when creating one). Just import sparql from osp.core.utils and use it to perform the queries. See the docs for more details on how to make queries (or just run help(sparql) on the Python interpreter).
• Improved the RDF import and export support. All import and export functionalities (including single cuds import/export and json-ld import/export) have been integrated into two utility functions: import_cuds and export_cuds. Check out the new tutorial on import and export capabilities for more details.
• Other changes and bug fixes (see the full changelog).

v3.5.2-beta

• Support for importing RDF files!
• Support for OWL restrictions and compositions. Now they have their own classes: Composition and Restriction. They can be accessed through the axioms attribute of an ontology class.
• Several performance improvements. CUDS objects instantiated within a with statement applied to a session are now directly created in such session.
• Some, mostly internal changes on how ontology entities are referenced in namespaces. User facing change: now, two entities cannot have the same label in the same ontology file. This prevents ambiguous references to entities.
• Other changes and bug fixes (see the full changelog).

v3.5.1-beta

Minor sql and backwards compatibility fixes.

v3.5.0-beta

• Support for reasoning via Wrapper: https://gitlab.cc-asp.fraunhofer.de/simphony/wrappers/factpluspluswrapper
• Updated serialization and sql database schema
• Data is now internally stored using rdflib, serialization uses json-ld
• Updated docstring style to google-style
• Allow to set pico install dir via environment variable: OSP_ONTOLOGIES_DIR
• Allow to install dcat and foaf (experimental)
• Bugfixes